I dont know if its currently a feature at the moment, but is there a way to reference a past prompt and output, so that the model is context aware of this.
A problem I realized is that, implementing a large feature and asking the models to do it introduces a lot of problems that are in the long-term actually worse, in comparison to breaking down a feature step by step. And asking cursor to implement step by step a subproblem and breaking it down into pieces.
Once a step is finished, I realize that in between from setting up the problem scope to completing the first step, there were many prompts in between to solve this step. And, if lets say I wanted to move to the next step, I would want to refer this step from the initial feature prompt I mentioned. A lot of times I would assume that this prompt is now out of the context window.
So, is there currently a way where we could make a reference to this initial problem prompt, so that the model is context aware of the problem we are trying to solve after an initial problem is solved? Does the model even remember the initial prompt? I would assume at a certain context length, it will not remember the initial problem its trying to solve.
